docs(#5548): fix stale FULLSEND_DIR description in eval script

[fullsend-ai-coder]

Summary

• Updates stale FULLSEND_DIR comment in eval/run-functional.sh from "path to fullsend scaffold directory" to "path to the .fullsend configuration directory"
• Aligns phrasing with the standard established by PR docs(cli): update --fullsend-dir help text from stale layout phrasing #5424 across eval/scripts/run-fullsend.sh and docs/testing/functional-tests.md

Context

PR #5424 standardized the FULLSEND_DIR description across 11 files but missed eval/run-functional.sh. The review agent flagged the gap as a Low finding, which was approved past without a fix. This PR addresses that remaining occurrence.

Testing

• Verified grep -r 'scaffold directory' eval/ returns no results
• Verified both eval/run-functional.sh and eval/scripts/run-fullsend.sh now use consistent ".fullsend configuration directory" phrasing

---

Closes #5548

Post-script verification

• Branch is not main/master (agent/5548-fix-fullsend-dir-comment)
• Secret scan passed (gitleaks — 5e2089bf999d52dbab62fd034d0f76af8b8538d7..HEAD)
• PR body secret scan passed (gitleaks — no-git)
• Pre-commit hooks passed (authoritative run on runner)
• Tests ran inside sandbox

[fullsend-ai-review]

🤖 Finished Review · ✅ Success · Started 8:43 PM UTC · Completed 8:50 PM UTC
Commit: 7549290 · View workflow run →

[github-actions]

Site preview

Preview: https://c270e010-site.fullsend-ai.workers.dev

Commit: 75492905383643167d9598b3b07799881ac0c7b2

[codecov]

Codecov Report

✅ All modified and coverable lines are covered by tests.

📢 Thoughts on this report? Let us know!

[fullsend-ai-review]

Looks good to me

[fullsend-ai-retro]

🤖 Finished Retro · ✅ Success · Started 6:43 AM UTC · Completed 6:55 AM UTC
Commit: 7549290 · View workflow run →

[fullsend-ai-retro]

Retro: PR #5549 — docs(#5548): fix stale FULLSEND_DIR description in eval script

Verdict: Successful end-to-end automated workflow. No new proposals — all improvement opportunities map to existing open issues.

Timeline

This PR originated from the retro agent's analysis of PR #5424, which standardized FULLSEND_DIR descriptions across 11 files but missed eval/run-functional.sh. The retro agent filed issue #5548 with a clear problem statement, proposed fix, and validation criteria. The full pipeline then executed autonomously:

Time (UTC)	Event
20:30	Retro agent files issue #5548 with ready-for-triage
20:32–20:35	Triage agent categorizes (Low/Documentation), applies ready-to-code (run)
20:36–20:41	Code agent makes 1-line fix, opens PR #5549 (run)
20:42–20:50	Review agent approves with "Looks good to me", applies ready-for-merge (run)
+10h (06:36)	Human (rh-hemartin) approves
+10h (06:40)	PR merged

Autonomous pipeline time: ~20 min (issue filed → PR reviewed and approved). Zero rework iterations. The code agent produced a correct fix on the first attempt.

What went well

• Retro detection worked: The retro agent correctly identified a gap left by a prior PR and filed a well-scoped, actionable issue.
• Pipeline coordination: Triage→Code→Review handoffs were clean, with no errors or retries.
• Code quality: The code agent produced a single, correct commit that precisely matched the issue's proposed fix.
• Review correctness: The review agent classified the change as "mechanical/value-only" and correctly approved.

Evidence for existing issues (no new proposals filed)

• agents#301 (lightweight path for trivial/mechanical changes): The review agent classified PR docs(#5548): fix stale FULLSEND_DIR description in eval script #5549 as "mechanical/value-only" but still dispatched 3 sub-agents (correctness, style-conventions, intent-coherence) + a challenger. All returned zero actionable findings. The review took ~9 minutes including ~1 min sandbox creation. This PR is additional evidence that the triage classification is working but has no lighter execution path to invoke.

• agents#280 (confirmatory reasoning in APPROVE comments): The review agent posted "Looks good to me" with no reasoning. Its internal verdict included clear reasoning ("Clean approve — trivial mechanical fix that completes the terminology standardization started in PR docs(cli): update --fullsend-dir help text from stale layout phrasing #5424, precisely scoped to issue Fix stale FULLSEND_DIR description in eval/run-functional.sh missed by PR #5424 #5548's authorization") but this was not surfaced in the PR comment.

• agents#373 (skip irrelevant dimensions for docs-only PRs): Although this PR changes a shell script comment (not pure Markdown), the change is documentation-in-nature. The correctness sub-agent had zero surface area to analyze.